I don't think it's just parameters. For example, if there are some property settings, if you pass SPI injection, many hidden dependency injections, plus comments are not conducive to source learning, how do you artificially? 我认为不单单是参数这些,比如有一些属性的设置,如果通过SPI注入,很多隐似的依赖注入,加上注释是不是跟有利于源码学习,你们是怎么人为的呢?
竹子 <[email protected]> 于2018年12月19日周三 下午8:50写道： > I think so 竹子 邮箱[email protected] 签名由 网易邮箱大师 定制 On 12/19/2018 20:41, > yuhang xiu wrote: Hi, all Recently I have seen a few pr about doc. Several > of them have been merged. I think we have two issues to discuss: 1. What > kind of comments should we add? 2. We need to define a goal. Regarding > question one, I personally think that we should: 1. Explain each param of > the method 2. Explain the special usage of some classes. For example, when > adding the doc of SerializableClassRegistry, we need to doc that this class > only works when using fst and kryo serialization. 3. We need to use @link > or @code for the association when the class (method) is associated with > other classes (methods). Regarding question two, I don't have a idea. > Perhaps doc coverage is 40% is a good choice. What do you think? yuhang xiu > <[email protected]> 于2018年12月18日周二 下午4:21写道： > Hello, Dave. > > (2) Have > any committers reached out to the users who had trouble with code >> > comments to confirm these efforts will meet their concerns? >> > > I used > to talk to them, but I forgot to ask them about what kind of docs > are > better. > > But don't worry, I still have their contact details. > I will > communicate with them and hope that they will bring their opinions > here > to better work for developers who add comments. >
